MAGNETISM AND
MAGNETIC MATERIALS
Edited by
Springer Science+Business Media, LLC
1962
MAGNETIC MATERIALS
Papers presented at the Conference on Magnetism and Magnetic Materials, Phoenix, Arizona, November 13-16, 1961
Sponsored by the American Institute of Electrical Engineers and the American Institute of Physics
Cooperating Societies:
Office of Naval Research
ISBN 978-1-4899-6193-8 ISBN 978-1-4899-6391-8 (eBook) DOI 10.1007/978-1-4899-6391-8
Copyright © 1962 Springer Science+Business Media New York Originally published by the American Institute of Physics, New York in 1962 Softcover reprint of the hardcover Ist edition 1962
This book, or parts thereof, may not be reproduced in any form without permission.
American Institute of Electrical Engineers Headquarters 33 West 39th Street, New York 18, New York
CONTENTS
Prologue ............... . . . . . . . . . . . . . . • • •• 1019'
General Session The State of the Art of Magnetic Memories ..................... Q. W. Simkins 1020 High Magnetic Field Research. . . . . . . . . . . . . . . . . . . . . . . . . . . Benjamin Lax 1025 Modification of Spin Screw Structure due to Anisotropy Energy and Applied Magnetic Field . . . . . .
· . . . . . . . . . . . . . . . . . . . . . . ..... Takeo Nagamiya 1029 Some Magnetic First-Order Transitions. . . . . . . . D. S. Rodbell and C. P. Bean 1037 Superconducting Materials and High Magnetic Fields. . ....... J. E. Kunzler 1042
Magnetic Devices Threshold Properties of Partially Switched Ferrite Cores . . . . . . . V. T. Shahan and O. A. Gutwin 1049 Properties of Magnetic Films for Memory Systems ................. E. M. Bradley 1051 Magneto-Optically Measured High-Speed Switching of Sandwich Thin Film Elements. . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . ..... J. C. Suits and E. W. Pugh 1057 Magneto-Optic Readout for Computer Memories .......... R. L. Conger and J. L. Tomlinson 1059 Switching Properties of Multilayer Thin Film Structures. . . . . A. Kolk, L. Douglas, and G. Schrader 1061 Magnetostatic Interactions between Thin Magnetic Films. . Harrison W. Fuller and Donald L. Sullivan 1063 Permalloy-Sheet Transftu.xor-Array Memory . . . . . . . . . .. G. R. Briggs and J. W. Tuska 1065 Demonstration of Magnetic Domain-Wall Storage and Logic ............ J. M. Ballantyne 1067 Flux Reversal in Ferrite Cores under the Effect of a Transverse Field. . . . . . . . . . . . Kam Li 1069 Approximate Solution of the Equations of Magnetization Reversal by Coherent Rotation . . . . . . . .
· ........................... R. F. Elfant and F. J. Friedlaender 1071 All-Magnetic Logic Elements Using Strained Permalloy Wire .............. H. R. Irons 1073 A Permanent Magnet Twistor Memory Element of Improved Characteristics. . . . . . . . . . . . .
· ............ E. J. Alexander, J. C. McAlexander, L. H. Young, and R. J. Salhany 1075
Internal Fields Anisotropy of the Hyperfine Interaction in Magnetite ........ E. L. Boyd and J. C. Slonczewski 1077 *de Haas van Alphen Effect in Zinc Manganese Alloys ........ F. T. Hedgcock and W. B. Muir 1079 Nuclear Resonance in Ferromagnetic Alloys .......................... .
· . . . . . . . . . . . Toshimoto Kushida, A. H. Silver, Yoshitaka Koi, and Akira Tsujimura 1079 Crystalline Electric Fields in Spinel-Type Crystals ................... V. J. Folen 1084 Hyperfine Fields, Spin Orbit Coupling, and Nuclear Magnetic Moments of Rare-Earth Ions ..... .
· ............................ R. E. Watson and A. J. Freeman 1086 Magnetic Hyperfine Interaction and Electronic Relaxation in Sm3+ in EuIG. . . . . . . . . . . . . .
· ....................... M. E. Caspari, S. Frankel, and G. T. Wood 1089 The Atomic Moments and Hyperfine Fields in Fe,Ti and Fe,Zr ..... C. W. Kocher and P. J. Brown 1091 Internal Magnetic Fields in Nickel-Rich Nickel-Cobalt Alloys. . . . . . . . . . . . . . . . . . . .
· . . . . . . . . . . . . . . . . . . . . . Lawrence H. Bennett and Ralph L. Streever, Jr. 1093
Thin Films-I Inhomogeneous Coherent Magnetization Rotation in Thin Magnetic Films. K. D. Leaver and M. Prutton 1095 Effect of Substrate Cleanness on Permalloy Thin Films . . . . . . . . . . . . . . . John S. Lemke 1097 Support and Extension of the Rotational Model of Thin Film Magnetization . . . . . . . . . . . . .
· . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. E. Schwenker and T. R. Long 1099 Induced Magnetic Anisotropy of Evaporated Films Formed in a Magnetic Field ... Minoru Takahashi 1101
* Abstract only.
Angular Dispersion and its Relationship with Other Magnetic Permameters in Permalloy Films. · . . . . . . . . . . . . . . . . . . . . . . . . Robert W. Olmen and Sidney M. Rubens 1107
Anisotropy Sources for Electrodeposited Permalloy Films ....... M. Lauriente and J. Bagrowski 1109
Constriction of Hard Direction Hysteresis Loops in Thin Permalloy Films. . . . . . . . S. Middelhoek 1111
Bitter Patterns on Single-Crystal Thin Films of Iron and Nickel. H. Sato, R. S. Toth, and R. W. Astrue 1113
Electron Microscope Study of the Roughness of Permalloy Films Using Surface Replication . . A. Baltz 1115
A Theoretical Model for Partial Rotation . . . . . . . . . . . . . . . . . . . . . . . H. Thomas 1117
Factors Influencing Coercive Force Values in Sputtered Permalloy Films. . . . . . . . . . . . . . . · ........................... A. ]. Noreika and M. H. Francombe 1119
Spin Configurations and Anisotropy
The Nature of One-Ion Models of the Ferrimagnetic Anisotropy .......... Peter]. Wojtowicz 1121
Magnetic Structure Work at the Nuclear Center of Grenoble. . . . . . . . . . . . . . . . . . . . · . . . . . E. F. Bertaut, A. Delapalme, F. Forrat, G. Roult, F. de Bergevin, and R. Pauthenet 1123
Neutron Diffraction Study of Magnetic Ordering in Thulium . . . . . . . . . . . . . . . . . . . . · . . . . . . . . . . . . . . W. C. Koehler, ]. W. Cable, E. O. Wollan, and M. K. Wilkinson 1124
*Temperature Dependence of the Magnetocrystalline Anisotropy of Face-Centered Cubic Cobalt . . . . · ..................................... D. S. Rodbell i126
Magnetoelectric Effects in Antiferromagnetics .............. G. T. Rado and V. ]. Folen 1126
Observation of Antisymmetric Exchange Interaction in Yttrium Orthoferrite . D. Treves and S. Alexander 1133
Neutron Diffraction Studies on Europium Metal ..... C. E. Olsen, N. G. Nereson, and G. P. Arnold 1135
Appearance of a Weak Ferromagnetism in Fine Particles of Antiferromagnetic Materials. . . . . . . . · . . . . . . . . . . . . . . . . . . W. ]. Schuele and V. D. Deetscreek 1136
*Magnetic Structure of Manganese Chromite. . . . . . . . . . . . L. M. Corliss and]. M. Hastings 1138
Lattice Theory of Spin Configuration . . . . . . . . . . . . . . . . . . . . . . . . E. F. Bertaut 1138
Magnetic Transitions in Cubic Spinels ........ N. Menyuk, A. Wold, D. Rogers, and K. Dwight 1144
Thin Films-2
Ferromagnetic Resonances in Thin Films ................ D. Chen and A. H. Morrish 1146
Stress Effects on the Magnetic Properties of Evaporated Single-Crystal Nickel Films . . ]. F. Freedman 1148
Stratification in Thin Permalloy Films. .
Electrodeposition of Magnetic Materials .
Rotatable Anisotropy in Composite Films.
. R. J. Prosen, ]. O. Holmen, B. E. Gran, and T. J. Cebulla 1150
. . . . . . . . . . . . . 1. W. Wolf 1152
. ]. M. Lommel and C. D. Graham, Jr. 1160
Unidirectional Hysteresis in Thin Permalloy Films. . W. D. Doyle, ]. E. Rudisill, and S. Shtrikman 1162
Ferromagnetic Resonance in Single-Crystal Nickel Films. M. Pomerantz, ]. F. Freedman, and J. C. Suits 1164
Isotropic Stress Measurements in Permalloy Films ........... G. P. Weiss and D. O. Smith 1166
Hall Effect Determination of Planar Stress in Ferromagnetic Films. . . . . . . . . . . . R. L. Coren 1168
Magnetoelastic Behavior of Thin Ferromagnetic Films as a Function of Composition. . . . . . . . . . · ............................ E. N. Mitchell and G. 1. Lykken 1170
Magnetization
*Statistical Mechanics of Ferromagnetism. . . . . . . . . . . Herbert B. Callen 1172
Magnetization of Localized States in Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . · .. P. A. Wolff, P. W. Anderson, A. M. Clogston, B. T. Matthias, M. Peter, and H. ]. Williams 1173
Pyromagnetic Test of Spin Wave Theory in Metallic Nickel . . Emerson W. Pugh and Bernell E. Argyle 1178
NMR in Domains and Walls in Ferromagnetic CrBr, .
Temperature Dependence of YIG Magnetization.
Ferromagnetic Resonance in CrBr3
A. C. Gossard, V. Jaccarino, and J. P. Remeika 1187
. Irvin H. Solt, Jr. 1189
. ]. F. Dillon, Jr. 1191
(continued on page v)
Oxides-l
"Acoustic Losses in Ferromagnetic Insulators ..... R. C. LeCraw, E. G. Spencer, and E. 1. Gordon 1193
X-Ray and Magnetic Studies of CrO, Single Crystals. W. H. Cloud, D. S. Schreiber, and K. R. Babcock 1193
Substitutions of Divalent Transition Metal Ions in Yttrium Iron Garnet . . . . . . . . . . . . . . · . . . . . . . . . . . . . . . S. Geller, H. J. Williams, R. C. Sherwood, and G. P. Espinosa 1195
Cation-Cation Three-Membered Ring Formation. . . . . . . . . . . . . . . . John B. Goodenough 1197
Perminvar Characteristics of Nickel-Cadmium Ferrites with Small Additions of Cobalt and Molybdenum · . . . . . . . . . . . . . . . . . . . . . . . . H. Lessoff and A. P. Greifer 1199
Low Temperature Anisotropy of Manganese-Iron Ferrites . . . . . . . . . . . . . . Wilfred Palmer 1201
Some Superparamagnetic Properties of Fine Particle oFeOOH ............ A. W. Simpson 1203
*Rare-Earth Ruthenates ....... R. Aleonard, E. F. Bertaut, M. C. Montmory, and R. Pauthenet 1205
Fine-Grained Ferrites. II. Ni[_xZnxFe204 ..... W. W. Malinofsky, R. W. Babbitt, and G. C. Sands 1206
Vanadium Iron Oxides ........... " A. Wold, D. Rogers, R. J. Arnott, and N. Menyuk 1208
Preparation and Properties of Ferrospinels Containing NiH . . . . . . . . . . . . . . M. W. Shafer 1210
Soft Magnetic Materials Two Effects of Changes in Tension. .
Recent Developments in Soft Magnetic Alloys .
Textured 6.5% Silicon-Iron Alloy. . . . . . .
....... Osamu Yamada 1212
. ...... Edmond Adams 1214
*Small Angle X-Ray Scattering from MnS in Silicon Steels. . . . . . . . . . . . . . . . . . . . .
· . . . . . . . . . . . . . . . . . . . Bani R. Banerjee, R. E. Lenhart, and W. H. Robinson 1224
A Method of Magnetic AnnEaling of Vanadium Permendur-Type Alloys. R. E. Burket and D. M. Stewart 1224
Magnetic Anisotropy of the Demagnetized State. . . Jean-Claude Barbier and Bernadette Ferlin-Guion 1226
Grain-Size Effects in Oriented 48% Nickel-Iron Cores at 400 Cycles. . . . . . . . . . . . . . . . . · . . . . . . . . . . . . . . . . . . . . . . M. F. Littmann, E. S. Harris, and C. E. Ward 1228
Magnetic Properties of Tape-Wound Cores of High-Purity 3% Silicon-Iron with the (llO) [OOlJ Texture · . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. L. Walter 1230
Cooling Rate Effect on Initial Permeability of 4-79 Molybdenum Permalloy. . . . . Armand A. Lykens 1232
Rare Earths
Magnetocrystalline Anisotropy of Rare-Earth Iron Garnets . . . . . . . . . . . . . . R. F. Pearson 1236
*Neutron Magnetic Scattering from Rare-Earth Ions . . . M. Blume, A. J. Freeman, and R. E. Watson 1242
Ferromagnetic Resonance in Terbium-Doped Yttrium Iron Garnet . . . . . . . . . . . L. R. Walker 1243
Far Infrared Spectra of Magnetic Materials. . . . . . . . . . . . . . . . . . . . . . M. Tinkham 1248
Electron Paramagnetic Resonance of Trivalent Gadolinium in the Yttrium Gallium and Yttrium Aluminum Garnets. . . . . . . . . . . . . . . . . . . . . . . . . . . . L. Rimai and G. A. deMars 1254
High Temperature Susceptibility of Garnets: Exchange Interactions in YIG and LuIG . Peter J. Wojtowicz 1257
The Contribution of Rare-Earth Ions to the Anisotropy of Iron Garnets . . . . . . . . . . . . . . . . . . . . . . .. ....... B. A. Calhoun, M. J. Freiser, and R. F. Penoyer 1259
Devices and Phenomena
Properties of Reset Cores in Radar Pulse Transformers. .
Coaxial Ferrite Phase Shifter for High Power Applications. Slow-Wave uhf Ferrite Phase Shifters. . . .
. . . . . . . . . . Reuben Lee 1261
. A. S. Boxer and R. S. McCarter 1263
. N. G. Sakiotis and D. E. Allen 1265
Magnetodynamic Mode Ferrite Amplifier. .. . Roy W. Roberts, Bert A. Auld, and Robert R. Schell 1267
Efficient Frequency Doubling from Ferrites at the 100-Watt Level.. .. A. S. Risley and I. Kaufman 1269
A Miniaturized Ferrimagnetic High Power Coaxial Duplexer-Limiter . . . . . . J. Clark and J. Brown 1270
* Abstract only.
Nonlinear Effects in Ferrite-Filled Reduced-Size Waveguide . . . . . . . . . . . . . . . . . . . . · . . . . . . . . . . . . . . ... . . . . . F. S. Hickernell, B. H. Auten, and N. G. Sakiotis 1272
Millimeter Wave Parametric Amplification with Antiferromagnetic Materials. . R. A. Moore 1274
A Microwave Magnetic Microscope ......................... R. F. Soohoo 1276
Sampling Magnetometer Based on the Hall Effect . . . . . . . . . . . .. ... H. H. Wieder 1278
New Type of Flux-Gate Magnetometer ..................... William A. Geyger 1280
Waveform of the Time Rate of Change of Total Flux for Minimum Core Loss ............ .
· . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H. L. Schenk and F. J. Young 1281
Low-Frequency Losses and Domain Boundary Movements in Silicon Iron ..... Daniel A. Wycklendt 1283
Iron Losses in Elliptically Rotating Fields .............. R. D. Strattan and F. J. Young 1285
Antiferromagnetism and Resonance
Ferromagnetic Resonance Magnon Distribution in Yttrium Iron Gamet . . . . . . . . . . . . . . . · . . . . . . . . . . . . . . . . T. J. Matcovich, H. S. Belson, N. Goldberg, and C. W. Haas 1287
Pulsed Critical Field Measurements in Magnetic Systems . . . . . . Simon Foner and Shou-Ling Hou 1289
Magnetic Susceptibility and Magnetostriction of CoO, MnO, and NiO . T. R. McGuire and W. A. Crapo 1291
Impurity Ion Effects in the Ferrimagnetic Resonance of Ordered Lithium Ferrite . . . . . . . . . . . · ...................... A. D. Schnitzler, V. J. Folen, and G. T. Rado 1293
Microwave Resonance Linewidth in Single Crystals of Cobalt-Substituted Manganese Ferrite. . . . . . · . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. W. Teale 1295
On the Possibility of Obtaining Large Amplitude Resonance in Very Thin Ferrimagnetic Disks . . . . . · .................................. F. R. Morgenthaler 1297
Ferromagnetic Resonance Loss in Lithium Ferrite as a Function of Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. T. Denton and E. G. Spencer 1300
Antiferromagnetic Resonance in MnTiO •............... J. J. Stickler and G. S. Heller 1302
Antiferromagnetic and Paramagnetic Resonance in CuF2·2H20 . . . M. Peter and T. Moriya 1304
dc Effects in Ferromagnetic Resonance in Thin Ferrite Films . . . . . . . . . W. Heinz and L. Silber 1306
Permanent Magnets and Micromagnetics
Failure of the Local-Field Concept for Hysteresis Calculations. . . . . . . . William Fuller Brown, Jr. 1308
The Formation of Monocrystalline Alnico Magnets by Secondary Recrystalization Methods. . . . . . . · .................. E. Steinort, E. R. Cronk, S. J. Garvin, and H. Tiderman 1310
The Preisach Diagram and Interaction Fields for Assemblies of y-Fe203 Particles. . . . . . . G. Bate 1313
Saturation Magnetization of Swagd Mn-AI. . . . . . . . . . ....... M. A. Bohlmann 1315
Studies of High Coercivity Cobalt-Phosphorous Electrodeposits ... J. S. Sallo and J. M. Carr 1316
Remanent Torque Studies in Polycrystalline BaFe120,.. . . . . P. J. Flanders and S. Shtrikman 1318
Permanent Magnetic Properties of Iron-Cobalt-Phosphides. . . . . . . . . . . . . . . . . . . . . · . . . . . . . . . . . . K. J. de Vos, W. A. J. J. Velge, M. G. van der Steeg, and H. Zijlstra 1320
The Effect of Angular Variations in Field on Fine-Particle Remanence .. Eric D. Daniel and R. Noble 1322
Possibility of Domain Wall Nucleation by Thermal Agitation. . . . . . . . . . . . Amikam Aharoni 1324
Study of Particle Arrangements and Magnetic Domains on the Surface of Permanent Magnets . . . . . · . . . . . . . . . . . . . K. J. Kronenberg 1326
Exchange Anisotropy-A Review .... . . . . . . . W. H. Meiklejohn 1328
Alloys and Compounds
Alloys of the First Transition Series with Pd and Pt ........... S. J. Pickart and R. Nathans 1336
Structural and Magnetic Properties of Copper-Substituted Manganese-Aluminum Alloys. . . . . . . . · . . . . . . . . . . . . . . . . . . . . . . . . . . Makoto Sugihara and Ichiro Tsuboya 1338
Neutron Diffraction Investigations of Ferromagnetic Palladium and Iron Group Alloys. . . . . . . . . · .............. J. W. Cable, E. O. Wollan, W. C. Koehler, and M. K. Wilkinson 1340
Magnetic Properties of Cr.S. in Chromium Sulfides . . . . . . . . . . . . . . . . . . . . . . . . · . . . . . . . . . . . . . . . . . . K. Dwight, R. W. Germann, N. Menyuk, and A. Wold 1341
(continued on page vii)
CONTE~TS (continued)
Anomalous Magnetic Moments and Transformations in the Ordered Alloy FeRh . . . . . . . . . . . · . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. S. Kouvel and C. C. Hartelius 1343
Age-Hardened Gold Permalloy and Gold Perminvar ......... E. A. Nesbitt and E. M. Gyorgy 1345
New Modified Mn 2Sb Compositions Showing Exchange Inversion. . . . . . . . . . . . . . . . . . · . . . . .. T. A. Bither, P. H. L. Walter, W. H. Cloud, T. J. Swoboda, and P. E. Bierstedt 1346
Magnetic Characteristics of Hydrogenated Holmium .......... Y. Kubota and W. E. Wallace 1348
Magnetic Susceptibility and Internal Friction of Tetragonal Manganese-Copper Alloys Containing 70 Percent Manganese. . . . . . . . . . . . . . ..... A. E. Schwaneke and]. W. Jensen 1350
Magnetic Susceptibilities and Exchange Effects in Four Organic Free Radicals . . . . . . . . . . . . · . . . . . . . . . . . . . . . . J. H. Burgess, R. S. Rhodes, M. Mandel, and A. S. Edelstein 1352
Paramagnetic Behavior of Iron-Rich Iron-Vanadium Alloys. . . . . . . . . . . . . . . . . . . . . · . . . .. .......... Sigurds Arajs, R. V. Colvin, Henry Chessin, and J. M. Peck 135,)
Magnetic Moment of Co-Cu Solid Solutions with 40 to 85% Cu. . . . . . . . . . . . . . E. Kneller 1355
Magnetic Structures of Mn2As and Mn2Sb07Aso3 . . . . . . A. E. Austin, E. Adelson, and W. H. Cloud 1356
Oxides-2 and Crystals
Investigations of Spin-Wave Interactions by the Parallel Pumping Technique ............ . · . . . . . . . . . . . . . . . . . . . . . . . . . . .. J. J. Green and E. Schlbmann 1358
Resonance Properties of Single-Crystal Hexagonal Ferrites. . . . . . . . . . . . . . . C. R. Buffler 1360
Growth of Yttrium Iron Gamet on a Seed from a Molten Salt Solution. . . . . . . . . . . . . . . . · . . . . . . . . . . . . . . . . . . . . . R. A. Laudise, R. C. Linares, and E. F. Dearborn 1362
Cobalt Ferrite Crystal Growth from the Ternary Flux System Na 20-CoO-Fe203 ........... . · . . . . . . . . . . . . . . . . . . . . . . . . . W. Kunnmann, A. Wold, and E. Banks 1364
Hexagonal Ferrites for Use at X- to V-Band Frequencies. . . . . . . . . . . . . . . . . . . . . . · . . . . . . . . . . . . . . . . . . . . . G. P. Rodrigue, J. E. Pippin, and M. E. Wallace 1366
High-Power Characteristics of Single-Crystal Ferrites with Planar Anisotropy . . . . Samuel Dixon, Jr. 1368
Nickel Aluminum Gallium Ferrites for Use at High Signal Levels.. J. W. Nielsen and J. E. Zneimer 1370
Subsidiary Absorption Effects in Ferrimagnetics ............ J. H. Saunders and J. J. Green 1372
Magnetic Properties of Yttrium-Gadolinium-Aluminum-Iron Garnets ... E. A. Maguire and J. ]. Green 1373
Temperature Stable Microwave Hybrid Garnets ...... Gordon R. Harrison and L. R. Hodges, Jr. 1375
Spin Wave Excitation in Planar Ferrites . . . . . . . . . . . . . Isidore Bady and Ernst Schlbmann 1377
Effect of Mechanical, Thermal, and Chemical Treatment of the Ferrimagnetic Resonance Linewidth of Lithium Ferrite Crystals . . . . . J. W. Nielsen, D. A. Lepore, J. Zneimer, and G. B. Townsend 1379
Growth of Single-Crystal Hexagonal Ferrites Containing Zn . . . . . . . . . . . . . . . . . . . . · . . . . Arthur Tauber, R. O. Savage, R. J. Gambino, and C. G. Whinfrey 1381
Author Index . 1383
Subject Index. . . . 1387
P. A. Albert J. F. Dillon S. Foner
D. B. Cachelin ]. F. Cubbage Mrs. D. L. Fresh D. L. Fresh
]. H. Crawford, Jr.
General Conference Chairman
J. A. Krumhansl ]. A. Osborn T. O. Paine
Committee Technical Program
Chairmen
Local Committee P. B. Myers, Chairman
F. R. Gleason C. S. Graniere L. ]. Ittenbach T. H. Kemp J. M. Lommel
Exhibits John Leslie Whitlock Associates
Publications Committee ]. A. Osborn, Chairman
1. S. Jacobs J. A. Krumhansl
Magnetic Materials Literature Digest
H. Rubenstein W. L. Shevel M. T. Weiss
J. B. Picone ]. D. Raile N. P. Rowley A. ]. Weslowski
]. S. Smart
J. C. Slonczewski, Editor W. Palmer, Assistant Editor W. H. Meiklejohn, Committee Chairman
]. E. Goldman C. ]. Kriessman J. A. Krumhansl
A. C. Beiler E. Both R. M. Bozorth F. G. Brockman H. Brooks H. B. Callen R. A. Chegwidden
AlEE Subcommittee on Magnetics L. R. Bickford, Jr., Chairman
E. A. Gaugler ]. E. Goldman D. M. Grimes C. L. Hogan 1. S. Jacobs C. ]. Kriessman V. E. Legg
Sponsoring Society Representatives American Institute of Electrical Engineers
R. S. Gardner American Institute of Physics
H. C. Wolfe
Cooperating Society Representatives The Institute of Radio Engineers
T. N. Anderson The Metallurgical Society of A. 1. M. E.
P. A. Albert Office of Naval Research
M. A. Garstens
M. F. Littman L. R. Maxwell W. H. Meiklejohn W. Morrill ]. A. Osborn T. O. Paine G. T. Rado
Journal of
Applied Physics Supplement to Volume 33, Number 3 March, 1962
"Let Thy Words Be Few"
Ecclesiastes 5:2
T HOSE venturesome magnetikers who made the trek west found Phoenix with the perfect weather
of the winter southwest: clear skies and invigorating air. For this the local Chamber of Commerce can thank their formidable ally C. L. Hogan who persuasively guided the Conference party to a very civilized frontier indeed. And now, as every year, the many who worked to make the Conference interesting and lively have asked themselves, "Was it a good meeting; was all the work worthwhile?" This is something that those attend­ ing the Conference must decide, but in this case many are both witness and jury as a large portion of those attending the meeting also participated in some part of the program machinery. A reasonable estimate would be that more than 150 people were required to ready the Conference, including nearly 100 referees. As this meet­ ing is run by volunteers (some in the army sense), it is a many-body system whose state is briefly observable but once a year. Although this sometimes seems like organi­ zational anarchy, it is in fact just this wide participation each year of new groups with fresh points of view that keep the Conference lively.
As the Conference has evolved its committees have experimented with ways in which it might be of greater service to those in the field of magnetism. This year the 1960 Magnetic Materials Literature Digest! prepared with scholarly devotion by J. Slonczewski, W. Meikle­ john, and their associates, was printed with Conference support and distributed to registrants.
In previous years we have attempted to communicate some of the excitement of science to high school students in the Conference city through speakers and a special high school day arranged in conjunction with the meet-
1 Copies at $2.00 each may be obtained from the American Institute of Physics, 335 East 45th Street, New York 17, New York.
ings. This year the Conference through its Phoenix colleagues (and some fine high school science teachers) sponsored student exhibits at the meeting which demon­ strated magnetic phenomena.
The variety of topics covered in the sessions was so large as to make a review here inappropriate. One sub­ ject was explored rather thoroughly by a number of authors: the nonlinear compressibility of contributed papers. While many, no doubt, experienced the sensa­ tion of being fitted to a procrustean bed, authors have made it easier to meet the restrictive time and proceed­ ings format each year. Hopefully because it is realized that these requirements are necessitated by the desire for an early appearance of the proceedings, the large size of the resulting volume, and economics. The issu­ ance of the Proceedings as a supplement of the Journal of Applied Physics and the essential help of its Publica­ tions Staff would have been impossible had the Confer­ ence not only been self-supporting but, additionally, been able to subsidize the Proceedings. The means which have been sought for Conference financial support then have been essential to its success: the exhibits, the ad­ vertising, and the bound version of the Proceedings by Plenum Press this year.
Thoughtful contributions and willing hands were many. The impersonal listing of Committee members elsewhere gives little idea of the long hours and late nights needed for these preparations. In particular, the Program Committee, under I. Jacobs and F. Luborsky, and the Local Committee (and the hardworking wives) headed by P. Myers. The editorial, refereeing gymkhana this year was successfully completed due to the generous participation of the many scientists and engineers who acted as referees and the willing labors of my colleagues on the Publication Committee.
J. A. OSBOR~ Editor
1019
JOURXAL OF APPLIED PHYSICS SlTPPLEMENT TO VOL. ,~3, NO.3 MARCH, 1962
General Session L. R. BICKFORD, Chairman
The State of the Art of Magnetic Memories Q. W. SIMKINS
Development Laboratories, Components Division, International Business Machines Corporation, Poughkeepsie, New York
A broad range of magnetic memories finds extensive use in today's data processing equipment. The most significant factors in evaluating a memory for a given application are reliability, cost, size, speed, and func­ tion. Ferrite core memories with capacities of 103 to 10· bits and with cycle times as short as 0.5 !,sec are in use. Recent developments in this area, including partial switching, 2 cores/bit, and new core fabrication techniques, will be discussed.
There have been extensive development efforts recently in many forms of magnetic metal film devices. Many geometries, modes of operation, fabrication techniques, substrate materials, and film compositions are being used. A comparison of some of the resulting devices will be made, and the effects of the factors listed above will be discussed. The application of metal film memory devices will be considered and a com­ parison drawn with ferrite devices.
Techniques for achieving special purpose functions such as nondestructive readout, read-only and associa­ tive memory using magnetic elements, will also be described and compared.
T HE pre-eminent position of magnetic memories in today's computer technology hardly need be
pointed out. While the field of magnetic surface record­ ing on tapes, disks, drums, cards, etc. is of much interest, my discussion will be limited to the field of electronically addressable magnetic memories. Memo­ ries of this sort cover a wide range of size, speed, and application. Registers and buffers storing as few as 1000 bits have successfully exploited magnetic tech­ niques, while today's most powerful data processors have one or more random access memories in the million-bit category.
Figure 1 shows the size and speed of a number of high performance memories. Complete read-write cycle times, including address decoding, are used, except as indicated. This chart is far from complete and lists only representative high-speed memories. For many applications, other parameters such as cost, reliability, size or power dissipation may be more important.
Ferrite core memories covering a wide range of speeds and sizes are available. A number of thin film memories are also represented in Fig. 1.
Most of the larger ferrite core memories employ coincident-current read and write,],2 and are often called bit-organized or three-dimensional memories. Many of the smaller and faster memories, including nearly all those using metal film devices, have coin­ cident-current write but non-coincident read and are called word-organized or two-dimensional memories.
In the remainder of this paper, I shall discuss the ferrite and metal film technologies, including recent device and memory system developments. Finally ex­ tensions of the art will be considered, with emphasis on the external circuits and the transmission lines re­ quired to link these circuits to the memory elements.
1 J. W. Forrester, J. App!. Phys. 22, 44-48 (1951). 2 J. A. Rajchman, Proc. Inst. Radio Engrs. 41, 1407-1421
(1953).
Since about 1955, ferrite cores have been the standard random access memory element. Many factors have contributed to the increased performance of ferrite core memories in terms of speed, size, reliability, and reduction in cost. Among the most significant have been the reduction in size of the cores and improvements in the manufacturing, automatic testing, and assembly techniques. Most of the newer memories employ 30/50 or 19/30 (inside diameter/outside diameter in mils) size cores and recent work indicates the feasibility of pressing3 or extruding and cutting4 13/21 cores. Smaller cores permit several things: a reduction in the drive requirements; closer packing, hence shorter drive and sense lines for a given memory size; lower drive line impedances and back voltages, simplifying driver design; and improved surface-to-volume ratios, giving better thermal characteristics. Some of these character­ istics are illustrated in Table I. The effect of core size on the array characteristics is shown in Table II.
2.0
o THIN FILM MEMORIES x FERRITE CORE MEMORIES
x x
MEMORY CAPACITY (BITS)
FIG. 1. Characteristics of existing high-performance memories.
3 M. H. Cook and E. C. Schuenzel, "A diminutive ferrite memory core," Electronics Division Fall Meeting, Am. Ceramic Soc., October 27, 1961.
4 W. L. Shevel, Jr., J. M. Brownlow, O. A. Gutwin, and K. R. Grebe, "New ferrite core arrays for large capacity storage," Symposium on Large Capacity Memory Techniques for Com­ puting Systems, Washington, D. C., May 23.1961.
1020
THE ART OF MAGNETIC MEMORIES 1021
Ferrite apertured plates in which many bits can be stored in a single pressed piece have also been devel­ oped. 5 Although the successful operation of memories employing such plates has been reported, they are not extensively used.
An interesting core plane assembly technique which permits partial machine wiring of memory planes is shown in Fig. 2. With this device all the X wires in a plane are fed through hollow needles and then simul­ taneously threaded through the cores. Y wires are similarly machine threaded before hand wiring of the Z and sense lines.
While three-dimensional memories are, in general, more economical,6 particularly in the larger sizes, two­ dimensional core memories do offer a number of ad­ vantages for high-speed operation. First, since very large read currents may be used, fast reading and high output signals may be obtained. Second, partial switch­ ing techniques8 may also be employed. Several schemes have been developed for efficient operation in the partial switching mode. For example, in the 2 core/bit arrange­ ment9 ,IO bipolar output signals are obtained by the differential connection of the sense line. This arrange­ ment has the further advantage that the drive line loading is independent of the stored information. The speed advantage of partial switching is shown in the plot of Fig. 3. This plot, which is especially useful in evaluating devices for 2-D applications, is obtained by the repeated switching of a core from its saturation state to a partially switched state by applying pulse fields of a given width and of increasing amplitude. Between each point obtained on the curve, the core is reset with a strong read pulse. Partial switching also reduces the core heating problem and permits the use of lower coercivity and less square core material. There are a number of problems, however, associated with partial switching of ferrite cores, including drive tolerances, in-
TABLE L FerriTe core charaCTeristics.
Half select Core size He current T,w S/V
(mils) (oersteds) (rna) (.u sec) (in.-l)
50/80 1.5 400 1.5 220 30/50 1.7 250 1.0 360 30/50 3.6 600 0.4 360 19/30 3.6 400 0.4 640 13/21 3,6 240 0.4 1000
5 J. A. Rajchman, Proc. Inst. Radio Engrs, 45, 325-334 (1957). 6 An interesting technique for reducing the number of diodes
required in a 2-D memory by utilizing the minority carrier storage effect is described by Melmed and Shevlin,7
7 A, Melmed and R. Shevlin, IRE Trans, on Electronic Com­ puters, Ee-8, No, 4, 474-478 (1959).
8 R. E. McMahon, "Impulse switching of ferrites," Solid State Circuits Conference, February, 1959, Digest of Technical Papers, pp, 16-17.
9 C. J. Quartly, Elec. Eng. 31, 756--758 (1959). 10 W. H. Rhodes, L. A, Russell, F. E. Sakalay, and R. M.
Whalen, IBM J. Research Develop. 4, No.2, 189-196 (1960).
TABLE II. Memory transmission line characteristics.
Density (bits/inch) l/to Memory elements B W Zo (bits/nsec)
Ferrite cores: 50/80 10 10 180 40 30/50 16 16 150 60 19/30 28 28 140 80 13/21 45 45
Flat thin film" 50 5 ",20 ",300
a These data are taken or inferred from reference 18.
creased line impedance, and a time dependent threshold effect. The latter is to be discussed in a paper by Shahan and Gutwinll at this conference.
The drive and sense lines through the cores of a very large high-density array behave like transmission lines. These lines, which may be as long as 70 ft, have a characteristic impedance of about 150 ohms and may have delays of up to 200 nsec. Considerable care must be taken to circumvent the effect of line delays and to avoid the coupling of energy from one drive line to another and from drive to sense line. Measures taken include the systematic placement of wires through cores, transposition between planes, the inversion of alternate core planes to reduce inductive coupling loops, the use of rectangular sense and inhibit segments, the accurate termination of all long lines, staggering of X and Y drivers, and staggering of the inhibit drivers and sense strobes corresponding to the X and Y line delays.12
Another development of major importance in large ferrite core memories is the load-sharing switch.13 ,14 By means of a core decoding switch that is transformer
FIG. 2. Machine wiring of core plane.
11 V. T. Shahan, and O. A. Gutwin, J. App!. Phys. 33, 1049 (1962), this issue.
12 C. A. Allen, E. D. Councill, and G. D. Bruce, Electronics 34, 68-71 (May 12, 1961).
13 G. Constantine, Jr., IBM J. Research Develop. 2, 204--211 (1958).
14 N. G. Vogl, "A new load-sharing matrix switch," 1961 Inter­ national Solid State Circuits Conference, February, 1961, Digest of Technical Papers, pp. 104--105.
1022 Q. W. SIMKINS
o 2 4 6 8 9 "IRITE FIELO (OERSTEOS)
coupled to the drive lines, the required drive power may be shared among a number of transistors. Since the addressing of a memory is arbitrary and programming restrictions are generally unacceptable, the memory must be capable of repeatedly interrogating a single address. By means of the load-sharing switch, the maximum power requirement on a single driver is greatly reduced, although the average drive power is unchanged. This makes possible the use of high-speed, medium-power transistors as memory core drivers.
While in limited use to date, memories employing thin magnetic metal film devices,15-19 have received much attention. Thin film memories are potentially attractive due to their very fast switching, excellent thermal properties and bulk fabrication. Most memories of this type involve the deposition of a thin film
BIT LINE
WORD LINE
FIG. 4. Thin film memory operation-parallel mode. ----
15 A. V. Pohm and S. M. Rubens, Proc. of the Eastern Joint Computer Conference, December, 1956, p. 120.
16 J. I. Raffel, J. App!. Phys. 30, 608-618 (1959). 17 E. E. Bittman, 1959 Solid State Circuits Conference, Digest
of Technical Papers, pp. 22-23. 18 J. 1. Raffel. T. S. Crowther, A. H. Anderson, and T. O.
Herndon, Proc. Inst. Radio Engrs. 49, No.1, 155-164 (1961). 19 E. M. Bradley, J. Brit. Inst. Radio Engrs. 20, No. 10, 765-
784 (October, 1960); Electronics 33,78-81 (September, 1960).
FIG. 3. Partial switching characteristics of a ferrite core.
10
(1000-2000 A) of nickel-iron (usually the non-mag­ netrostrictive 81% Ni-19% Fe composition) on a flat substrate of glass, mica, or metal. The films may be deposited by plating, by vacuum evaporation, or by sputtering in the presence of an orienting field to achieve uniaxial anisotropy. Individual spots of either circular or rectangular shape with dimensions varying from 10 to 100 mils may be either etched or obtained by evapora­ tion through a mask. Bradley has reported work19 in which the individual bits in a large sheet were defined only by the crossings of the drive lines. This arrange­ ment has the added advantage of simplifying the regis­ tration problem.
Thin film memories have been operated in both paralleP7 and perpendicular18 drive modes. In the par­ allel mode, one arrangement of which is illustrated in Fig. 4, the word and bit fields in a 2-D array are applied parallel but off the easy axis to assure fast rotational switching. If sufficiently good film properties could be obtained, a memory in this mode could be operated in the bit-organized or 3-D fashion, but to date this has not been feasible. Dietrich and Proebster have shown20
that irreversible flux changes or "creeping" occur for fields much less than the wall motion coercive force He for fields applied at 45°, while this does not occur for fields applied in the easy direction. This phenomenon makes the realization of a parallel mode thin film memory very difficult.
The perpendicular drive mode thin film memory is illustrated in Fig. 5. Here the word field is applied in the hard direction, rotating the magnetization through 90° either in the clockwise or counterclockwise direction depending on the initial state of magnetization which corresponds to the stored information. This rotation induces either a positive or a negative signal in the sense line which runs in the easy direction (perpen-
20 W. E. Proebster and W. Dietrich, International Solid Stale Circuits Conference, February, 1961, Digest of Technical Papers, pp.66--67.
THE ART OF MAGNETIC MEMORIES 1023
dicular to the word line). A small bit field in the easy direction applied at the end of the word pulse deter­ mines the new rest state of the magnetization and there­ fore controls the writing process. A variation on the perpendicular drive mode involves a slight rotation of the easy axis, so that there is a preferred direction after reading and a unipolar bit pulse may be used.
Individual films of the type described above have been switched in as little as 1 nsec.21 In principle, large numbers of bits can be fabricated simultaneously (the current state of the art seems to be about 100018 ,19).
Recent progress in obtaining greater film uniformity is significant in this regard; however, uniformity, particu­ larly edge effects and dispersion and variation in the easy axis presently limit plane sizes to about 10 square inches or less. Another major problem in memories of this type is the relatively low signal level obtained. A simple calculation shows that with a 40-mil spot, 1000 A thick, the maximum signal attainable is about 10-4
v-}J.sec even with perfect coupling of the drive and sense lines. A typical high-density film array might have a I-mv signal level with 20-nsec switching. This implies that film devices of this geometry must be switched very rapidly, and therefore fast rising drive fields must be used. The difficulty of generating such fields, together with the noise problem in the sense line, has limited the field of application of thin film memories to high-speed, low-capacity memories. Since these devices do not have a closed flux path, they are also relatively sensitive to fields from all sources. This must be taken into account in their design, and usually dictates the minimum spacing of the bits in a plane.
The sandwich or dual film geometry, in which a second film with an easy axis parallel to the first is placed over the first with the stripline wiring be­ tween, reduces the demagnetization problem some­ what but introduces additional fabrication and uni­ formity problems.18
A number of film memory elements employing other geometries have also been described. While most of the twistor22 memories use a thin Permalloy tape wrapped on a copper wire, they can be fabricated by plating on a wire23 and are considered here as film type memory devices. Both the twistor, which has a helical easy axis, and the magnetic rod24 memory element, which has a longitudinal easy axis, have an open flux path. This is shown in Fig. 6. In the twistor, the word and bit drive fields are orthogonal (both at 45° to the easy axis). The rod, which has a high (140e) coercivity to reduce the self-demagnetization, makes use of parallel drive fields in the easy direction supplied by currents flowing in lO-turn solenoids.
21 W. E. Proebster, W. Dietrich, and P. Wolf, IBM J. Research Develop. 4, No.2, 189- 196 (1960).
22 A. H. Bobeck, Bell System Tech. J. 1319- 1340 (1957), 23 S. J. Schwartz and J. S. Sallo, IRE Trans. Oil Electronic Com­
puters, EC 8, No.4, 465- 469 (1959). 24 D. A. Meier, Proc. of the Electronic Components Conference,
May, 1960, pp. 122- 128.
WORO LINE ~DlINE
memoryoperation-per- pendicular mode.
SENSE~ STROBE
Cylindrical films25- 28 deposited either on metallic or glass substrates with a circumferential easy axis have also been reported (see Fig. 6). Memory elements of this type may be operated in either the parallel or perpendicular mode. An advantage of this geometry is the absence of a demagnetizing field in the static remanent state, permitting the use of thicker films with resultant higher signal levels. High-speed switching has also been reported with these devices, and recent work in our laboratory indicates that switching con­ stants as low as 0.01 oe-}J.sec can be achieved in this geometry.
The switching properties of magnetic memory ele­ ments play a major role in determining memory per­ formance; however, in many applications, particularly in the sub microsecond speed range, the associated logic, driver, and sense circuits and the interconnecting trans­ mission lines are controlling. Many film memory ele-
TWISTOR
ROD
FIG. 6. Cylindrical geometry film memory devices.
25 T. R. Long, J. App!. Phys. 31, 123S- 124S (1960). 26 T. R. Long, "A discussion of electrodeposited cylindrical film
memory elements," Electrochemical Society !\h·t'ling, Oc(ol>u, 1961.
27 G. R. Hoffman, J. A. Turner, and T . Kilburn , ]. Brit. Ill st. Radio Engrs. 20, No . 1, 31-36 (1960).
28 G. Rostky, Electronic Desi~n 9, 26-27 (August 1961).
1024 Q. W. SIMKINS
men is switch in the rise time of the drive field; thus, not only are very wide band pass drive and sense circuits required, but in addition the array must be so designed that tolerable loss, distortion, and cross coupling will occur in the transmission paths. If the signal delay is comparable to the rise time, it is generally necessary to terminate the lines in their characteristic impedances to prevent excessive ringing and distortion; this increases the required drive power. Strip trans­ mission lines, in which the width-to-separation ratio is high, are attractive because of their low impedance, well-defined field pattern, and relative insensitivity to inductive coupling. These considerations have led to the use of very thin glass18 or metaP9 substrates for flat film memories. In some cases, loss and distortion in these lines may limit the size of memories or at least of sense segments. The importance of high density memory element packaging is apparent from the foregoing. Com­ parative data for a number of memory elements are shown in Table II.
Much of the present research and development effort in the memory area is aimed at extending memory capacity or speed and at special function memories. Memory capacity limitations are primarily economic, thus any technique for reducing the cost of memory elements or reducing the drive requirements is beneficial in this area.
As described above, many of the film devices switch very rapidly; consequently, memory speed is largely determined by the band width and delay of the drivers, sense amplifiers, transmission lines, and logic stages in the regeneration loop. In two-dimensional read-write memories, cycle time is often limited by the sense circuits' recovery time from writing. This effect may be limited by reducing the coupling from the bit to sense line,19 by termination of the lines, or by write noise cancellation in a bridge circuit.18 ,29
Another means of achieving high speed in a memory system is the use of nondestructive read out (NDRO). Since there are usually several times as many read and regenerate operations as there are write operations, a fast NDRO read, "slow" write memory has many potential applications. The potential speed advantage of an NDRO memory is considerable; not only does it save the switching time associated with the read operation, but also it saves the time required for the regenerative sense loop to function and avoids the troublesome sense circuit write recovery problem. A
29 W. B. Gaunt and D. C. Weller, "A 12-kilobit, 5-microsecond twistor variable store," 1961 International Solid-State Circuits Conference, February, 1961, Digest of Technical Papers, pp. 106- 107.
number of NDRO memory techniques have been shown to work, including cross field interrogation of ferrite cores,30 permeability sensing of cores,31,32 ac sensing of cores,33 multipath ferrite devices,34 bias restoration of cores,35 hard film restoration of soft films in film pairs,36,37 and less than 90° rotation of thin film memory elements. Most of these techniques, particularly those relying on the reversible switching of flux, are troubled by low output signals and small margins.
Other types of special function memory include read­ only and associative memories. Read-only memories are a special class of NDRO memories in which the informa­ tion must be altered mechanically. Magnetic read-only memories fall into two classes; those which involve linear coupling elements such as Kilburn's peg memory,38 and those which involve nonlinear coupling elements such as the permanent magnet twistor memory.39 The advantage of the nonlinear coupling element is a reduction in noise sensitivity. An associative memory35 is not normally addressed in the conventional manner but rather a word is called out based on a match of portions of the stored words. A means of achieving this type of memory with magnetic elements based on the bias restored core NDRO cell is described by Kiseda et al.35
In summary, compact, highly reliable ferrite core memories are used widely today in data processing systems. Present work in the field, including a substan­ tial effort in the development of a variety of thin film devices, promises higher speed and perhaps more versa­ tile memories for future use.
30 R. M. Tillman, IRE Trans. on Electronic Computers EC-9, 323-328 (1960).
31 G. H. Perry and S. J. Widdows, "Low coercive-force ferrite ring cores for a fast non-destructively read store," International Solid-State Circuits Conference, February, 1960, Digest of Tech­ nical Papers, pp. 58-59.
32 W. L. Shevel, Jr. and O. A. Gutwin, "Partial switching, non­ destructive-readout storage systems," International Solid-State Circuits Conference, February, 1960, Digest of Technical Papers, pp.62-63.
33 R. E. McMahon, "ac and impulse switching techniques for fixed, random access and analog memory use," International Solid-State Circuits Conference, February, 1961, Digest of Technical Papers, pp. 68-69.
34 J. A. Rajchman and A. W. Lo, Proc. Inst. Radio Engrs. 44, 321-322 (1956).
35 J. R. Kiseda, H. E. Petersen, W. C. Seelbach, and M. Teig, IBM J. Research Develop. 5, No.2, 106-121 (1961).
36 L. J. Oakland and T. D. Rossing, J. App!. Phys. 30, 54S-55S (1959).
37 R. J. Petschauer and R. D. Turnquist, "A nondestructive readout film memory," Proceedings of the Western Joint Com­ puter Conference, May, 1961, pp. 411-425.
381. L. Auerbach, Proc. Inst. Radio Engrs. 49, No.1, 330-331 (1961).
39 W. A. Barrett, F. B. Humphrey, J. A. Ruff, and H. L. Stadler, IRE Trans. on Electronic Computers EC-IO, No.3, 451-461 (1961).
JOURNAL OF APPLIEll PHYSICS SUPPLEMEXT TO VOL. 33, NO.3 MARCH, 1962
High Magnetic Field Research
National Magnet Laboratory,* Massachusetts Institute of Technology, Cambridge 39, Massachusetts and Lincoln Laboratory,t Massachusetts Institute of Technology, Lexington 73, Massachusetts
The use of high magnetic fields as a research tool for a wide variety of physical phenomena is clearly recognized today and the subject of an International Symposium at Cambridge this fall. Another important step in promoting research with the help of large magnetic fields has been the sponsorship of the M. I. T. National Magnet Laboratory by the Air Force. This paper will review the highlights of the conference which included papers on research in plasma physics, low temperatures, solid state and the latest developments for generating high magnetic fields. The plans and objectives of the National Magnet Laboratory and descrip­ tion of the physical facilities will be presented. In addition, a brief review will be given of a number of experi­ ments already performed in the existing Magnet Laboratory at M. I. T.
INTRODUCTION
SCIENCE, like many other creative efforts of man­ kind, is characterized by fashions and trends. In the
proper course of evolution certain ideas and tools achieve that state of maturity which inevitably makes them ripe for fruitful exploitation. The early period of this century was marked by the intensive productivity of spectroscopy and related atomic theory. During the late 1920's and 30's research was strongly influenced by quantum theory and nuclear physics stimulated by accelerators, which even today are important tools in high energy physics. After World War II solid-state physics emerged as a major field of science due to the impact of microwave and low temperature techniques and the advances in materials science. Today we stand on the threshold of a new era which may become known as the decade of high magnetic field research. Within the last year two events may be regarded as milestones which herald the coming of this era. The first Inter­ national Conference on High Magnetic Fields, which was attended by more than 700 conferees at the Massa­ chusetts Institute of Technology, indicated a growing enthusiasm and interest in this subject. A more con­ crete evidence of progress and enhanced future activity was the beginning of the construction of the National Magnet Laboratory.
The conference which was sponsored by the Solid State Sciences Division of the Air Force Office of Scien­ tific Research had two major themes. The first of these was centered specifically around the art of generating high magnetic fields and involved research, design, and development of solenoids of different types. The second focused attention upon the physical research which utilizes these solenoids and which covered a broad spectrum including solid-state physics, low tempera­ tures, plasma physics, fusion research, and biomagnetics. It is not possible in a short review such as this to give a proper coverage of the conference, consequently only the highlights of some of the areas will be discussed.
* Supported by the Air Force through the Air Force Office of Scientific Research.
t Operated with support from the U. S. Army, Navy, and Air Force.
HIGH MAGNETIC FIELDS
Magnetic fields represent an old tool for research. However, only a few attempts beyond the 10 000 gauss range have been made until relatively recently. In general there are now four major approaches to the generation of high magnetic fields, each probably com­ plementary to the other; namely, pulsed magnets, water-cooled coils, cryogenic magnets and supercon­ ducting coils.
Pulsed Fields
Pulsed systems are relatively inexpensive and easy to construct from condenser banks, a switch and some type of wound or rigid coil design. These can generate useful fields from 100000 gauss to perhaps 500 000 gauss depending on the particular problem to be pursued. Usually the measurement techniques are fairly intricate with such transient systems. Nevertheless, with inge­ nuity pulsed magnets have progressed beyond the stage of an exploratory instrument. One of the early applica­ tions of pulsed fields was made by Kapitza1 who has pioneered in the measurement of the magnetoresistance of a number of materials. This type of study has been revived by the Harvard group2 and others and the work on metals has been reported by Olsen3 at this conference. The beautiful experiments of Schoenberg and co­ workers on the de Haas van Alphen! effects in copper and other metals was discussed by Priestley. a More modern experiments on cyclotron resonance5 and millimeter wave spin resonance6 have been carried out at Lincoln Laboratory and were reported by Foner.3 Other ex­ amples of pulsed field experiments are those in magne­ tism, the susceptibility measurements of paramagnetic materials by Stevenson,3 and the novel spin-flop experi­ ments on antiferromagnets by Jacobs.a Pulsed fields are
1 P. Kapitza, Proc. Roy. Soc. (London) A123, 292 (1929). 2 H. P. Furth and R. W. Waniek, Phys. Rev. 104,343 (1956). 3 Papers presented at the 1961 International High Magnetic
Field Conference will appear in a copy of the Proceedings to be published by the Technology Press (1962).
4 D. Shoenberg, The Fermi Surface (John Wiley & Sons, Inc., New York, 1960), p. 74.
5 B. Lax, J. G. Mavroides, H.]. Zeiger, and R. W. Keyes, Phys. Rev. 122,31 (1961).
6 S. Foner, Phys. Rev. 107,683 (1957); J. phys. radium 20, 336 (1959).
1025
1026 !1 F N J A 1\1 T l'\ L A X
also becoming very important for determining the tran­ sition fields of hard superconductors in the hundred thousand gauss range. Bubble chambers and other devices for analysis of high energy particles are now employing pulsed magnets. The production of shock waves in plasmas by large pulsed fields for producing high temperatures was described by Kolb3 of the Naval Research Laboratory. High energy storage in inductors was considered by Carruthers3 of the Atomic Energy Research Establishment, England. Another interesting phenomenon that was described by Furth3 during the proceedi~gs was the plastic deformation of soft metals such as copper and aluminum by the large magnetic forces at the center of a pulsed coil. Finally when pulsed magnets are discussed the technique of implosion devel­ oped by Fowler3 must be included. With this technique he is able to generate momentarily fields in excess of 10 megagauss.
Water-Cooled Magnets
The second method for generating high magnetic fields involves the use of water-cooled coils. Such sys­ tems at approximately 2-megawatt level, capable of generating fields of the order of 100000 gauss now exist in a number of laboratories throughout the world. In this country, in addition to the present magnet labora­ tory at M. I. T., there is one at the Naval Research Laboratory, the University of California, N. A. S. A. Lewis Research Center, and the one at Bell Telephone Laboratories. Others are being planned for the near future. In England there now are three; Oxford, Cam­ bridge, and at the Radar Research Establishment in Great Malvern. There is a two-megawatt installation at Leiden, Holland. There is also one in Wrocklaw, Poland, and two in Japan, one at Tokyo University and one at Tohoku. Larger installations, of course, are being built at M. 1. T., which I shall discuss shortly, and one of comparable magnitude is being planned at the Lebedev Institute in Moscow. The possibility also exists that in England a third such large facility will be built.
The areas of research that have been first exploited by the use of water-cooled magnets are spectroscopy, low temperature, and magnetism. More recently, solid-state physicists and plasma physicists have also become good customers of high magnetic fields. A number of the speakers, who discussed various fields of research that are being carried on with magnetic fields, made an excellent case for the steady-state water-cooled mag­ nets. One of these was Kurtj3 from Oxford who presented cogent arguments why larger magnetic fields of the order of 200 000 gauss or more would be invaluable in adiabatic nuclear polarization and adiabatic nuclear cooling. Similarly, Professor Bloembergen3 of Harvard, who talked about spin resonance, clearly demonstrated that larger magnetic fields would permit greater sensi­ tivity in nuclear resonance, higher resolution in para­ magnetic resonance, and in particular, would permit the
observation of resonance in paramagnetic and antiferro­ magnetic systems where the natural frequencies due to crystalline fields or internal effective fields occur in the far infrared region of the spectrum. This latter possi­ bility has already been partially exploited and demon­ strated by the pulsed technique in a number of such resonance experiments discussed by Foner.3 John Galt3
of Bell Telephone Laboratories reviewed cyclotron resonance experiments in solids and showed that larger magnetic fields would be extremely helpful in exploring the absorption curves of metals at microwave frequen­ cies whereby the anomalous skin effect could be mini­ mized by the magnetic field. Cyclotron resonance of ions at high magnetic fields were experimentally illustrated by Buchsbaum3 of Bell Telephone Laboratories where the high fields permitted studies of the linewidth and resolution of multi-ion systems which otherwise could not be observed. In the solid-state area we also showed that high continuous fields was extremely important for the study of magneto-optical studies of semicon­ ductors, metals and magnetic materials. Water-cooled magnets have also been applied advantageously to measurement of the magnetization of rare earths, which was reviewed by W. Henry. Oscillatory effects in solids in high fields was considered by Kahn of the National Bureau of Standards. Another area where steady mag­ netic fields are important for the study of physical properties, is in plasma physics. Professor Allis3 of M. I. T. demonstrated from theoretical considerations how large magnetic fields permit the study of plasma waves over a wide range of the parameters involved and in which a variety of phenomena, including cyclotron resonance, Alfven waves, whistler modes, and magneto­ hydrodynamic waves could be profitably investigated. The problem of ambi-polar diffusion in plasmas, which is not only of fundamental interest, but is also inevitably encountered in any plasma studies was discussed by Lehnert3 of the Royal Institute of Technology in Stock­ holm. In addition, the role of high magnetic fields in producing magnetohydrodynamic phenomena such as "wakes" was considered by Sears3 of Cornell University.
The practical problems of generating magnetic fields with water-cooled systems was one of the important areas considered in this conference. Recent advances on water-cooled systems was discussed by Professor Bitter3
of M. I. T. who described the basic problems involved in the design of coils from consideration of heat trans­ fer, mechanical strength of materials and power avail­ ability. He predicted that water-cooled systems could ultimately reach the one-half million gauss mark with powers of the order of SO megawatts on a continuous basis or with a low-duty cycle. In addition to the paper presented by Professor Bitter on the ultimate limits to be achieved with water-cooled coils, a number of other people discussed different systems and different power supplies utilized in energizing the coils. Parkinson3 of the Radar Research Establishment discussed a battery system which for a short period of the order of 15 min or
HIGH MAGNETIC FIELD RESEARCH 1027
so was capable of achieving fields as large as 128 000 gauss. Another interesting system which was presented was the one at N. A. S. A. Lewis Research Center, which utilized a homopolar generator3 at low voltage and high currents and a relatively simple water-cooled copper magnet capable of achieving 100 000 gauss quite readily. Perhaps one of the most attractive systems discussed during the conference was that presented by Adkins3 of Cambridge who made a good case for a silicon rectifier system capable of supplying two megawatts with good stability and low ripple if operated during the night. The attractive part was that the cost was much less than that of some of the other systems, which balances some of the disadvantages. In addition to these systems, various other large magnet installations involved with the fusion work were presented, such as the Stellarator at Princeton by Mills,3 the large mirror machine at the Lawrence Radiation Laboratory, California, by Coens­ gen.3 Other large water-cooled magnets for fusion will be presented in fair detail in the Proceedings of the Conference.
Cryogenic and Superconducting Magnets
The next major class of magnets that received a great deal of attention during the conference were the cryo­ genic systems, and this may be divided into two major categories, superconducting and nonsuperconducting. In the latter category several efforts were reported among which the most advanced was that of Laquer3 at Los Alamos on hydrogen-cooled copper magnets which have attained fields of the order of 60000 gauss or more. Aluminum magnets were also discussed by Taylor3 of Lawrence Radiation Laboratory and the group at the National Bureau of Standards in Boulder, Colorado. One of the most ambitious undertakings in this regard was the neon-cooled system3 to be used with the homopolar generator by the group at N. A. S. A., Cleveland. An extremely attractive system used by the Philips group at Eindhoven involved a liquid nitrogen precooled magnet3 which was then pulsed by 100-kw generator to provide long pulses of the order of seconds.
The real excitement of the conference came from the superconducting magnets which will be discussed in detail by Dr. Kunzler3 in a succeeding paper. Without stealing his thunder, let me just merely state that already four different groups at Bell Telephone Labora­ tories, Westinghouse, Lincoln Laboratory, and Atomics International have super conducting magnets of either niobium tin or niobium zirconium in excess of 50 000 gauss. An interesting paper on the field analysis of superconducting coils was given by Gauster3 of the Oak Ridge National Laboratory. In some respects the super­ conducting magnet appears to be the poor-man's instru­ ment from the initial capital cost and a useful one pro­ vided he has available helium and cryogenic techniques in his laboratory. There is no question that supercon­ ducting magnets will play an exceedingly important role
in high field magnet research in the future. One can speculate on its particular role in high field magnet research in the future. One can speculate on its partic­ ular role at the moment, however, like many new developments, it will in all likelihood complement the other systems and in individual situations will offer advantages, and in some cases such as large volume magnets, it will be the only practical solution.
NATIONAL MAGNET LABORATORY
The National Magnet Laboratory is sponsored by the Air Force Office of Scientific Research and operated by M. 1. T. In spirit and operation it will be similar to the Brookhaven National Laboratory, but on a smaller scale. The research activities will primarily be in the solid state and low temperature areas and will include an active program of research and development of magnets. The construction of the Laboratory has just begun and all the electrical equipment, motor genera­ tors, bus bars and other hardware are on order. The system will utilize large water-cooled magnets which will be capable of achieving 250 000 gauss in a one-inch working internal diameter. In order to achieve this field with copper coils, a total of 8 megawatts of continuous power will be supplied by four 2-megawatt dc generators operating from two synchronous motors, each with a fly wheel. The fly wheels will provide a pulse capability at four times the normal power or 32 megawatts for several seconds. In addition, we will be able to sweep the mag­ netic field and operate any of four of the machines individually or in suitable parallel-series combinations. The current fluctuations will be maintained at less than 0.015%. The ripple voltage at the fundamental fre­ quency of 6 cps will be less than 0.02%.
The floor plan for the operating portion of the Magnet Laboratory is shown in Fig. 1, including the motor generator, sets, etc., which will be located in a new build­ ing with special foundations to minimize vibrations. There will be eight special magnet rooms, each with a coil for solid state and other experiments which require
OTHER SERVICES
ALBANY STREET
RIVER 70' ---I fl n
FIG. 1. Floor plan of the National Magnet Laboratory now under ~onstruction, showing the location of two motor generator sets WIth ten magnet rooms to house water-cooled solenoids for high field research.
1028 BENJAMIN LAX
o SOURCE MOVING AWAY FROM ABSORBER
500 1000 1500 2000 2500 3000
VELOCITY (1000 = 1.12 mm/sec)
FIG. 2. Mossbauer effect in stainless steel in a longitudinal field of 64 kgauss showing a single Zeeman line. (After N. Blum.)
large space. The building will allow for expansion in the event that it cannot accommodate all the participants ~nticipated in the near future. The bus bars and the water lines will run in the basement underneath the magnet rooms and the power will be automatically switched from one coil to another from a central control room. In addition to the water-cooled magnets, there will be pulsed magnets and ultimately a combination of large superconducting magnets with the water-cooled magnets which will permit a wide variety of possibilities. The present aim of the Laboratory is to achieve fields of the order of 250000 gauss at the 8-megawatt level but eventually to exceed these and aim for the 300 000 to 400000 gauss range with the fly wheel pulse system, and with the combination of superconducting and water­ cooled magnets.
Present Laboratory and Research Program We have remodeled the present magnet laboratory
which has a power supply of 1. 7 megawatts and a variety of magnets capable of providing fields up to 126 000 gauss. The most convenient magnet has been the one with the two-inch bore, capable of achieving 90 000 gauss and has been used for a large number of experi­ ments. The list of experiments that have been performed by the staff of the National Magnet Laboratory and graduate and post-doctoral students includes a variety of magneto-optical effects in semiconductors, para­ magnetic and antiferromagnetic materials, and mag­ netic properties of antiferromagnetics, which were dis­ cussed by Dr. Foners at this meeting, and measurements of critical fields in hard superconductors. Among these I also want to mention the Mossbauer effect at high fields which is of particular pertinence to this conference. Cooperative experiments with members of Lincoln Laboratory, faculty and staff of other laboratories at M. I. T. and those in the local area are listed as follows: magnetotunneling in semiconducting diodes, magneto­ reflection experiments in metals, cyclotron resonance in diamond, magnetic susceptibility measurements, and behavior of conducting fluids in magnetic fields.
About two years ago when Pound7 first reported his Mossbauer Zeeman effect in Fe57, it occurred to me that our high magnetic fields might possibly make an im­ portant contribution in this area. At that time, the plan was to use the external magnetic field to measure the g factor of both the excited and ground state of the nucleus and the internal field, with one set of measure­ ments at high fields. In the meantime, however, this was accomplished by Hanna and his group8 at Argonne with low field polarization experiments and combined with nuclear resonance. Nevertheless, Norman Blum at the National Magnet Laboratory proceeded with the experi­ ment and indeed the results which he obtained together with some more recent data in cooperation with Pro­ fessor Grodzins of M. 1. T. are beginning to give some interesting information about internal fields in these materials. Figure 2 shows the simple Zeeman spectrum obtained in Fe57, in this particular case in stainless steel, where it is necessary to use a large external field and indeed we obtained a single peak consistent with the theory. The object is to study the behavior of this peak as a function of magnetic field as shown in Fig. 3. From this plot of the Zeeman shift it turns out that with the knowledge of the g factors, it is possible to measure the actual field seen by the nucleus which is equal to the external field minus a hyperfine field. From this one can then measure the hyperfine constant. This experiment has now been carried out in stainless steel and also iron. The promising future for this type of experiment is that it can be done in many materials which have no internal fields and will permit the measurement of the spin factors for the ground and excited states of the nucleus as well as the hyperfine constants in these materials. In general it appears to be a very powerful technique for studying the internal field problem.
Figure 4 shows the results of the four-millimeter cyclotron resonance in diamond carried out by Rauch of Lincoln Laboratory. Some months ago he reported the
~ MOSSBAUER EFFECT :; "'_ 7
MAGNETIC FIELD (KILOGAUSS)
FIG. 3. Energy shift of Zeeman line with applied magnetic field in stainless steel. (After N. Blum.)
---- 7 R. V. Pound and G. A. Rebka, Jr., Phys. Rev. Letters 3, 439
(1959); 3, 554 (1959). 8 S. S. Hanna, ]. Heberle, C. Littlejohn, G. J. Perlow, R. S.
Preston, and D. H. Vincent, Phys. Rev. Letters 4, 177 (1960).
HIGH MAGNETIC FIELD RESEARCH 1029
first two peaks9 which he was unable to identify. By using monochromatic excitation he showed that the 1.07 mo peak was that due to the spin orbit split-off valence band. Hence from the 0.7 mo light hole it was possible to estimate the heavy hole to be 2.2 mo. At 70 kMc this required fields of the order of 60 kgauss. Hence the experiments were repeated with higher fields to obtain all three masses as shown.
CONCLUSIONS
In order to carry out the experiments to date, we have had to work on a two-shift basis and are already starting on a three-shift operation in order to accommodate additional experiments that are being planned for the existing magnet laboratory. It is expected that in about one and a half years the new laboratory will be in oper-
9 C. Rauch, Phys. Rev. Letters 7, 83 (1961).
FIG. 4. Cyclotron res­ onance in diamond at 70 kMc in high magnetic field. Peaks correspond to 0.7 ma light hole, 1.07 ma split-off hole, 2.2 ma heavy hole. (After C. Rauch.)
Ij~1 10 20 30 <10 50 eo lB8
MAGNETIC FIELD (kllo-H~ted)
ation and will permit many more researchers from other organizations as well as graduate students and visiting scientists to participate and perform experiments with high fields. Since the sponsorship comes from the Solid State Sciences Division of the Air Force Office of Scien­ tific Research, it is expected that the bulk of the pro­ gram will be in solid state. However, the new facilities will be made available to research workers of other fields as well, with the object of making this a truly National Laboratory and with an international flavor as well.
JOURNAL OF APPLIED PHYSICS SUPPLEMENT TO VOL. 33. NO.3 MARCH. 1962
Modification of Spin Screw Structure due to Anisotropy Energy and Applied Magnetic Field
TAKEO NAGAMIYA
Department of Physics, Osaka University, Osaka, Japan
Recent development in the theory of screw structure of spins and its modification due to anisotropy energy and applied magnetic field is reviewed, with reference to neutron diffraction work on rare-earth metals and MnAu2.
1. INTRODUCTION
T HE screw structure of spins, first predicted by Y oshimori1 and later by Villain2 and Kaplan,3 has
found its application in a number of examples. Rare­ earth metals with more than half-filled 4/ shells, ranging from Tb to Tm (Tb, Dy, Ho, Er, Tm), present intricate but interesting examples. For these metals magnetic measurements have been done by the people at Ames4 and neutron diffraction measurements by the people at Oak Ridge. 5 They all show ferromagnetism at low temperatures, not always a simple ferromagnetism in actuality, and apparently antiferromagnetism at intermediate temperatures. This apparent antiferro-
1 A. Yoshimori, J. Phys. Soc. Japan 14, 807 (1959); T. Nagamiya, J. phys. radium 20, 70 (1959)-Proc. International Conference on Magnetism, Grenoble, 1958.
2 J. Villian, J. Phys. Chern. Solids 11, 303 (1959). 3 T. A. Kaplan, Phys. Rev. 116, 888 (1959). 4 D. R. Behrendt, S. Legvold, and F. H. Spedding, Phys. Rev.
109, 1544 (1958); B. D. Rhodes, S. Legvold, and F. H. Spedding, Phys. Rev. 109, 1547 (1958); J. F. Elliott, S. Legvold, and F. H. Sped ding, Phys. Rev. 100, 1595 (1955); R. W. Green, S. Legvold, and F. H. Spedding, Phys. Rev. 122, 122 (1961).
5 M. K. Wilkinson, W. C. Koehler, E. O. Wollan, and J. W. Cable, J. App!. Phys. 32, 48S (1961); W. C. Koehler, J. App!. Phys. 32, 20S (1961); J. W. Cable, E. O. Wollan, W. C. Koehler, and M. K. Wilkinson, J. App!. Phys. 32, 49S (1961).
magnetism was actually found to be a screw in Dy and Ho (possibly also in Tb) and a sinusoidally varying spin arrangement in Er and Tm. It is very likely that the crystalline field anisotropy energy of the 4/ shell of these metal atoms has modified the simple screw arrangement of spins which these metals would have had in virtue of a certain characteristic exchange interaction if there were no anisotropy energy. A study of the crystalline field anisotropy energy and its effect on the spin arrangement has recently been done by Miwa and Y osida, 6 Elliott,7 and Kaplan,8 independ­ ently, and they succeeded in interpreting the magnetic structures of these metals.
A magnetic field acting on the crystal should also affect the magnetic structure. A magnetic field is, so to say, the source of the simplest anisotropy energy which can be varied at will. The effect of an applied field on the screw structure of spins has been studied at Saday
6 H. Miwa and K. Yosida, Proc. International Conference on Magnetism and Crystallography, Kyoto, 1961 (to be published).
7 R. J. Elliott, Proc. International Conference on Magnetism and Crystallography, Kyoto, 1961 (to be published); Phys. Rev. 124, 346 (1961).
8 T. A. Kaplan, Proc. International Conference on Magnetism and Crystallography, Kyoto, 1961 (to be published); Phys. Rev. 124, 329 (1961).
1030 TAKEO NAGAMIYA
by Herpin and Meriel9 with MnAu2 by neutron diffrac­ tion measurements. Rare-earth metals are also being investigated at Oak Ridge by Koehler.1° The corre­ sponding theory, applicable to simplest cases only at the present moment, has been developed by Herpin and Meriel,9 Enz,ll and the present writer.l2
It is the purpose of this memorandum to briefly review the information at present available and to outline the theories mentioned above.
2. OUTLINE OF THE THEORY OF SCREW STRUCTURE
Before going over to the main part of the present article, it might be helpful to briefly outline the theory of screw structure. Consider a set of layers of atoms whose spins are coupled ferromagnetically within each layer with an exchange constant J 0, between adjacent layers with J 1, between next-nearest-neighboring layers with J 2, and so on. Let the spins in the same layer be parallel and their direction be specified by an angle On, measured in one plane from a certain specified direction, where n is the number of the layer under consideration. Then the interaction energy can be written
E= -52 Ln [Jo+2h COS(On+1-0n) +2J2 COS(On+2-0n)+· .. J, (1)
where, for the sake of simplicity, the number of pairs interacting in the same way are supposed to be included in the exchange constants. If we put On = nq+const, we have
E= -N52J(q), where J (q) = J 0+ 2J 1 cosq+ 2J 2 cos2q+ ... , (2)
N being the number of layers. Thus, the minimum of the energy corresponds to the maximum of J(q), and if J(q) is the largest at q=O or 71" the system will show ferromagnetism or antiferromagnetism, respectively. If, however, J (q) is the largest at qo different from 0 and 71", the system will have a screw structure, in which the spin vectors rotate uniformly with an angle qo as one goes from layer to layer. Y oshimori has proved rigor­ ously that for crystals consisting of equivalent magnetic atoms-equivalent in the sense that the environment of every atom, as regards the magnetically interacting neighbors, is the same apart from translation-the screw structure represents the stable solution of the problem of minimum energy if J(q) is the largest at qo. The example he took was Mn02, for which he explained beautifully the neutron diffraction lines observed by
9 A. Herpin and P. Meriel, C. R. Acad. Sci. 250, 1450 (1960); preprint from Service de Physique du Solide et de Resonance Magnetique, Centre d'Etudes NucJeaires de SacJay.
10 w. C. Koehler, Proc. International Conference on Magnetism and Crystallography, Kyoto, 1961 (to be published).
11 U. Enz, Physica 26, 698 (1960); J. App!. Phys. 32, 22S (196l). 12 T. Nagamiya, K. Nagata, and Y. Kitano, Proc. International
Conference on Magnetism and Crystallography, Kyoto, 1961 (to be published); Progr. Theor. Phys., Kyoto (to be published).
Erickson,13 with qo=S7I"/7. His proof also applies to hexagonal close-packed structures if the screw axis is parallel to the hexagonal axis, which is the case for rare-earth metals with more than half-filled 41 shell. (If the screw axis is perpendicular, or in general oblique, to the hexagonal axis, the uniform rotation has to be modified in such a way that, corresponding to two atoms in each unit cell, the spin vectors of one sublattice is rotated by a certain angle with respect to those of the other sublattice.)
The discovery of screw structure was made through the following consideration. Several years ago, Nakamura and Nagai14 at Kyushu University were investigating the spin wave spectrum of MnF 2, which has the same crystal structure as Mn02. The spin wave frequency starts at a finite value at zero wave number, corresponding to the antiferromagnetic resonance frequency, but it goes down with increasing wave number qz along the z axis if the exchange constant between atoms neighboring along the c axis is greater than the exchange constant between a corner atom and a neighboring body-center atom. If the ratio of these exchange constants exceeds 1.2, the frequency becomes negative in a certain interval of qz. This suggested that for such a value of the ratio the simple antiferromagnetic structure known for MnF2 must be unstable and that instead a certain static spin wave must be realized. This idea of the present writer lead to Yoshimori's discovery of screw structure. It is noted in passing that a recent experiment by Owen15 of the ESR in MnF2 diluted with ZnF2 revealed that the ratio mentioned above is very small and is negative.
The neutron diffraction pattern to be obtained from a screw structure is as follows. As known by the theory of magnetic scattering of-neutrons, only the component of the spin vector parallel to the plane of reflection is effective for magnetic lines. In this plane, the component along one coordinate axis and that along the other coordinate axis give rise independently to the line intensity if the neutrons are unpolarized. Each com­ ponent oscillates sinusoidally in a screw structure so that the plus and minus phase differences between consecutive planes due to this sine wave, namely +qo and -qo, come into the scattered wave. Therefore the Bragg condition is modified as
k'- k±qo= K, (3)
where k' and k are, as usual, the wave vectors of the scattered and incident neutrons, respectively, and K is a reciprocal lattice vector. The direction of the vector qo represents the direction of the screw axis. The ordinary Bragg reflection lines are therefore .each split
13 R. A. Erickson, unpublished work. 14 T. Nakamura and O. Nagai, unpublished work; O. Nagai
and A. Yoshimori, Progr. Theoret. Phys. (Kyoto) 25, 595 (196l). 15 J. Owen, Proc. International Conference on Magnetism and
Crystallography, Kyoto 1961 (to be published); M. R. Brown, B. A. Coles, J. Owen, and R. W. H. Stevenson, Phys. Rev. Letters 7, 246 (1961).
MODIFICATION OF SPIN SCREW STRUCTURE 1031
into two, corresponding to ±qo, and by measuring the magnitude of the splitting one can determine the value of qo. It is interesting to note that the reflection lines with K=O can also be observed.
As can be seen from the above argument, the period of the screw has nothing to do with the lattice spacing. This is in fact a new feature characteristic of the screw structure. Furthermore, as long as only exchange inter­ actions are concerned, the plane in which the spins rotate is indeterminate. If this plane is perpendicular to the direction along which the rotation propagates, i.e., perpendicular to the direction of the screw axis, then the screw is called "proper." On the other hand, if this plane contains the screw axis, then the screw may rather be called "cycloidal." Now, when one takes into account the anisotropy energy or an applied field, the situation is changed in the following way. In the case where the anisotropy energy stabilizes the plane perpendicular to the screw axis, the screw is proper; when it stabilizes a plane which contains the screw axis, the screw is a cycloid. When there is a large uniaxial anisotropy energy which stabilizes the screw axis, as in the case of Er and Tm, the cycloid may happen to be squeezed to become a collinear array of spin vectors. If the anisotropy energy is such that it stabilizes a circular cone whose axis coincide with the screw axis, the spin vectors may rotate on a cone, giving rise to a ferro­ magnetic component along the screw axis, as in the low temperature phase of Ho and that of Er. For this anisotropy energy, it is also possible that the spin vectors rotate on a plane which is oblique or parallel to the spin axis, as in the intermediate temperature phase of Er, since then the spin vectors can point alternately near the upper cone and the lower cone and at the same time can lower the exchange energy by rotating on a plane. In the case of a proper screw, the anisotropy energy within the plane will modify the uniform rotation of the spin vectors. If this anisotropy energy has a six­ fold symmetry and is large enough, the spin structure will be one of the following four: (1) ferromagnetic, when the turn angle of the spin vectors in the original screw is less than 30°, as realized in the low temperature phase of Dy, (2) a screw of pitch six, when the turn angle is between 30° and 90°, (3) a screw of pitch three, when the turn angle is between 90° and 150°, (4) anti­ ferromagnetic, when the turn angle exceeds 150°. For moderately large anisotropy energy in the plane, the spin arrangement will be complex; for the case where the system is going to be ferromagnetic, the spin vectors will oscillate sinusoidally in the neighborhood of one of the easy axis, as one may guess from the result to be described later in the study of the effect of strong applied field.
The effect of an applied field is different depending on whether the anisotropy energy is small or large. In the case where a small anisotropy energy stabilizes the plane perpendicular to the screw axis and a magnetic field is applied in the same plane, the initial magnetiza-
tion corresponds to a small tilt of the spin vectors toward the field direction. Then, by increasing the field, one will find that the proper screw changes into a cycloid whose plane is perpendicular to the field, since the susceptibility of the cycloid is larger than that of the proper screw and, therefore, by converting the sys­ tem from proper screw to cycloid above a certain critical field, one gains more by the decrease in the energy of exchange and interaction with the magnetic field of the system than one loses by the increase of the anisotropy energy. This change is similar to that known in antiferromagnetic crystals. In this case, a further increase of the field will make the spin vectors point closer to the field direction and finally make all of them parallel to this direction. In the case where the anisot­ ropy energy confines the spin vectors firmly in the plane, the effect of the applied field is somewhat peculiar, as will be described later. The effect of the applied field for various cases of anisotropy energy and for finite temperature has not yet been studied fully.
The origin of the exchange interaction that gives rise to a screw structure is not yet fully understood. In compounds, such as Mn02 and FeCh, the ratio of the superexchange constants may happen to conform to the condition of stabilizing a screw structure. In MnAu2 and in rare-earth metals, indirect exchange interaction through s-d and s-f interaction, respectively, might be responsible for their screw structures; this interaction is, as known well, extends over a number of atomic distances and changes sign alternately as a function of distance. The general trend of the observed Neel temperature in the second half of the rare-earth series is in agreement with that expected from this inter­ action. 6,1,16
3. ANISOTROPY ENERGY AND MAGNETIC STRUCTURE IN RARE-EARTH METALS
Elliott1 and Miwa and Yosida6 have studied the crystalline field anisotropy energy for the second half of the rare-earth series. The high-temperature suscepti­ bilities of the metals give the Curie constants appro­ priate to an assembly of tripositive rare-earth ions. They have a well-defined total angular momentum J com­ posed of the total spin angular momentum S and the total orbital momentum L, Land S taking the maxi­ mum possible values in accordance with the Hund rules and pointing parallel to each other. Th